letter nature genetics • volume 27 • january 2001 55 A mechanism for exon skipping caused by nonsense or missense mutations in BRCA1 and other genes Hong-Xiang Liu 1,2 , Luca Cartegni 1 , Michael Q. Zhang 1 & Adrian R. Krainer 1 1 Cold Spring Harbor Laboratory, Cold Spring Harbor, New York, USA. 2 Present address: Xencor Inc., Pasadena, California, USA. Correspondence should be addressed to A.R.K. (e-mail: krainer@cshl.org). Point mutations can generate defective and sometimes harmful proteins. The nonsense-mediated mRNA decay (NMD) pathway minimizes the potential damage caused by nonsense muta- tions 1–4 . In-frame nonsense codons located at a minimum dis- tance upstream of the last exon-exon junction are recognized as premature termination codons (PTCs), targeting the mRNA for degradation. Some nonsense mutations cause skipping of one or more exons, presumably during pre-mRNA splicing in the nucleus; this phenomenon is termed nonsense-mediated altered splicing (NAS), and its underlying mechanism is unclear 1,2,5,6 . By analyzing NAS in BRCA1, w e show here that inappropriate exon skipping can be reproduced in vitro, and results from disruption of a splicing enhancer in the coding sequence. Enhancers can be disrupted by single nonsense, missense and translationally silent point mutations, without recognition of an open reading frame as such. These results argue against a nuclear reading- frame scanning mechanism for NAS. Coding-region single- nucleotide polymorphisms 7 (cSNPs) within exonic splicing enhancers or silencers may affect the patterns or efficiency of mRNA splicing, which may in turn cause phenotypic variability and variable penetrance of mutations elsewhere in a gene. Several models have been proposed to explain the link between nonsense mutations and RNA-processing alterations 1–6 . Pro- posed mechanisms for NMD include scanning of ORFs by a hypothetical nuclear machinery; nonsense-codon recognition during translocation of partially spliced mRNA through nuclear pores; and tagging of exon–exon junctions, immediately after nuclear splicing, by putative factors that are later recognized dur- ing the first round of cytoplasmic translation. These proposed mechanisms may also be relevant to NAS, which has additionally been attributed to disruption of sequences or secondary struc- tures involved in exon definition, to instability of the exon- included form of the mRNA due to NMD, or to RNA assay artifacts 1,2,6,8–11 . Besides nonsense mutations, some missense and silent mutations also cause skipping of constitutive exons 6,8,12,13 . Exon sequences comprise cis -acting elements that influence the use of flanking splice sites. For example, exonic splicing enhancers (ESEs) are present in constitutive or alternative exons of certain genes, and are required for efficient splicing of those exons 8–14 . The ESEs in pre-mRNAs are recognized by serine/arginine-rich (SR) proteins, a family of essential splicing factors that also regulate alternative splicing 15,16 . Each SR protein recognizes a distinct, albeit degenerate, functional sequence motif under splicing conditions 17–19 . Score matrices for four SR proteins, derived from their functional consensus sequences, were recently used to show that high-score SR protein motifs are enriched in exons, especially in regions corresponding to known, natural ESEs (refs. 17,19). An amber (TAG) nonsense mutation (Glu1694Ter), consisting of a G→T transversion at position 6 of exon 18 of the breast can- cer susceptibility gene BRCA1, causes inappropriate skipping of the entire constitutive exon in vivo 20 . This mutation was found in a family with eight cases of breast or ovarian cancer, and five other independent cases were reported in the BRCA1 Informa- tion Core Database. Skipping of exon 18 results in retention of the same reading frame and removal of 26 amino acids, disrupt- ing the first BRCT (for BRCA1 C terminus) domain of BRCA1 (ref. 20). How nonsense mutations cause exon skipping has been unclear. ESE disruption is one proposed mechanism 2,6,8–10 , although inactivation of known ESEs typically requires deletions or multiple substitutions. Because most characterized ESEs are purine-rich, and the Glu1694Ter mutation does not occur in a purine-rich segment, ESE inactivation was not favored as a mechanism 20 . But certain non-purine-rich seg- ments can enhance splic- ing 17,19,21,22 ; therefore, we tested whether this mutation in BRCA1 inactivates an ESE. We used SF2/ASF, SC35, SRp40 and SRp55 motif-scor- ing matrices 17,19 to analyze exon 18 of wild-type and mutant BRCA1. Multiple high-score motifs of each type are distrib- uted throughout this exon (Fig. 1). The mutation specifically disrupts the first of three high- ATGCTG AGTTTGTGTGTGAACGGACACTGAAATATTTTCTAGGAATTGCGGGAGGAAAATGGGTAGTTAGCTATTTCT 0 1 2 3 4 5 Fig. 1 High-score SR protein motifs in BRCA1 exon 18. We searched the 78-nt sequence of wild-type exon 18 with four nucleotide-frequency matrices derived from pools of functional enhancer sequences selected in vitro 17,19 . Motif scores refl ect the extent of matching to a degenerate consensus, and only the scores above the threshold for each SR protein are shown. High-score motifs are shown in black for SF2/ASF, dark gray for SC35, light gray for SRp40 and white for SRp55. The width of each bar reflects the length of the motif (6, 7 or 8 nt); the placement of each bar along the x axis, the position of a motif along the wild-type exon DNA sequence; and the height of the bar, the numerical score on the y axis. The thresholds and maximal values are different for each SR protein. The G at position 6 (wild type) is highlighted. The nonsense mutation that changes this G to a T only affects the fi rst SF2/ASF motif, reducing the score from 2.143 to 0.079 (below the threshold). © 2001 Nature Publishing Group http://genetics.nature.com © 2001 Nature Publishing Group http://genetics.nature.com